e19514 Background: Lower-intensity regimens, such as hypomethylating agents (HMAs) are better tolerated and more effective than standard chemotherapy in elderly patients( > 65 years old) with acute myeloid leukemia (AML). Scoring systems to predict clinical outcome in these patients are mainly based on cytogenetic assays and morphological response to therapy. Minimal residual disease (MRD) represents a powerful risk- stratification in younger AML patients but it remains poorly investigated in older patients, particularly in those undergone HMAs. In this subset, the survival benefit is not limited to patients achieving morphologic complete remission (CR). The clearance of leukemic blasts was never be valued as a predictive parameter of outcome in older patients. Thus, in this study we evaluated the most effective and predictive factor of outcome at the optimal time-point among leukemic blasts clearance by flow cytometry (FC), MRD and morphological response in older patients treated with HMAs. Methods: From april 2015 to may 2019, 76 older patients ( > 60 years old) with newly diagnosed AML were treated con azacitidine and decitabine. Among them, 56 completed at least six cycles but 31were properly monitored by FC. Log Clearance was defined as logarithmic ratio between the percentage of blast cells at specific time point and that at the diagnosis. Results: A total of 31 elderly patients with a median age 72 (60-85) were evaluated after 4(T4) and 6(T6) cycles of HMAs. A Median of 10 cycles were administered, equally distributed between azacitidine and decitabine. A higher frequency of CR (73.3%vs 20%) and MRD negativity (16.6% vs 0) was found at T6 compared to T4, thus T6 was chosen for the analysis. Performance status (PS > 2),Hgb < 10 g/dl, MRD positivity and a Log-reduction lower than 2 log predicted a significantly shorter DFS at univariate analysis. However, Log-reduction and PS only confirmed to predict a short DFS also at multivariate analysis, together with CRi (CR with incomplete recovery) and partial remission (PR). When the OS was investigated, only PS represented a strong indicator of prognosis at both univariate and multivariate analysis. Conclusions: Our results suggest that patients achieving 2 Log at least of blasts-reduction by FC had a longer DFS. Dynamic monitoring of blast –reduction stratified better and more widely than the absolute value of MRD in older AML patients. Having a high PS and achieving a CRi instead of a CR showed an unfavorable outcome in terms of both DFS and OS